ZOD Sequences
The ZOD Sequences dataset comprises 1473 high-quality, 20-second sequences with comprehensive sensor coverage. For an in-depth overview of the sensor configuration, please visit the Sensor setup section on the Home page.
Annotation Details
At present, the ZOD Sequences dataset is annotated only for a single keyframe, corresponding to the central camera frame within the sequence. These annotations strictly adhere to the Frames dataset style, encompassing annotations for Objects, Lanes, Traffic Signs, and Ego Road. For a thorough understanding of each annotation task, kindly refer to our Annotation page.
We plan to release full-sequence annotations for some tasks in the near future, so stay tuned!
Samples
Below, we showcase a selection of data samples from the dataset. These samples have been chosen to specifically highlight interesting and diverse traffic scenarios.